ME1 Citadel FPS Drops

@PilotGW  @NaaMei @Schachmatt123 

To get some info here, how big is the difference in the utilisation of your highest utilised CPU core between an area where the game runs "normally" and the areas where there are frame drops?

@holger1405 

Here are two screenshots from Edolus. As you can see, GPU usage drops significantly when facing the "bad spot". After turning to another direction, framerate goes up along with GPU usage at nearly 100%. Highest CPU usage for one core is at 49% while facing the bad spot, and 31% while facing another direction.

GPU is a GTX 970, CPU is a Core I7 2600k

@holger1405 

8700K, RTX 3080 / 1440p, 144Hz GSync monitor (FPS limited externally to 141) / All settings maxed

Same place as @Schachmatt123, I observe the following:

"Bad" scenario

Facing west towards the open map, FPS drops from 141 down to 53.

Usage of the highest CPU "core" is around 40% average, however it's variable between 35-45%. My CPU is hyperthreaded, and the highest "thread" usage on that core is around 70%, bouncing on occasion to the mid 80%, but never above that.

GPU usage is around 40% with the clock speed between 1250-1300MHz.

"Good" scenario

Facing east towards the map edge, FPS is stable at 141.

Usage of the highest CPU "core" is around 15-20% with the associated thread also lower.

GPU usage is around 45% (presumably as it's allowed to "spread it's wings" so to speak and render 141 FPS) with the clock speed between 1450-1700MHz (my 3080 maxes out at 1980MHz for reference).

So CPU "thread" usage definitely increases in my case, but that thread never maxes out. The rest of my CPU is sat mostly unutilised, which shouldn't be the case for a game released in 2021, remaster or not. Besides, in my tests with OG ME1, the CPU usage is roughly the same, without the associated FPS drops. GPU has very little bearing in my case because the 3080 is so powerful.

@Schachmatt123 

The CPU usage would still be to low for a bottleneck, so I assume that for you there is also one thread that is much more taxed?

@PilotGW 

---

PilotGW wrote:

So CPU "thread" usage definitely increases in my case, but that thread never maxes out. The rest of my CPU is sat mostly unutilised, which shouldn't be the case for a game released in 2021, remaster or not. Besides, in my tests with OG ME1, the CPU usage is roughly the same, without the associated FPS drops. GPU has very little bearing in my case because the 3080 is so powerful.

---

Yeah, my 3080 is bored as well. 😉
And you are right, it seems the game is using just one thread for the heavy lifting, that really should be not the case for a remaster. 

But 70% is still strange, I found one spot, at the area at the bridge before the Council Tower coming from the embassies, where the FPS drop to about 90 - 100 for me, but here this single thread goes well into the 90%, what I would expect for a bottleneck.

@holger1405 

Here are some screenshots of the fps drops I encountered on the citadel. There are two really noticeable locations.

Location 1: right out the citadel embassy.

Location 2: one of the elevator going down to the wards (I think) with a Turian chilling on the bench outside. 

From what I noticed, its really hard to tell what the CPU usage is, because for me all cores are bouncing from as low as 15% usage to as high as 70% usage in both normal FPS scenario and when the FPS drops. Although when the FPS is normal CPU usage is more stable. I would say the average is around 40% when stable. No cores were maxed out even during the FPS dips. 

I'm running the game with Vsync off on an average 4k monitor without no G-sync, Freesync, adaptive sync etc. My CPU is i7-6700HQ and GPUis RTX 2080Ti hooked up through a graphics amplifier (sorry rtx 2080ti for abusing you xd, you deserve better than to be caged like this).

@NaaMei 

• Press "STRG + ALT + Del" to open the Task Manager.
• Open "More Details" if it is not already set this way.
• Click on the "Performance" Tab
• At the bottom of the Performance Tab > Click on "Open Resource Monitor"
• Choose the "CPU" tab.
• Now all threads of your CPU should be visible in graphs.

One thread should be clearly more taxed in game.

@holger1405 

You can see in both my screenshots the Core Temp window showing the cpu usage 🙂

In the "bad scenario" one core is at 49%, the second highest at only 15%. In the "good scenario" CPU usage is more evenly spreaded.

Edit: Screenshots added again

Thanks @Schachmatt123 !

I was more curious about the single thread performance, but if the entire core is at 49% it is save to asume that one thread of this core is also at least close to 90%.

So, this is clearly a CPU bottleneck.  

My goodness, I had heard Dr. Saleon's ship is a black hole of performance but this is even worse than I thought.

@Femmeworth 

Yes, this is bad. 😮

But this is not a CPU bottleneck as it seems, so there is something else going on. 

@Femmeworth  Am I reading this correctly as in 79.94%(!) of CPU time in user mode is used up by GetDenuvoTimeTicketRequest in the ME1 exe? Because that looks... pretty out of the ordinary, considering that is DRM functionality.

@EA_Atic  Any updates on this issue? There isn't really a lot of communication from upstream as to which of the issues in these forums are being tracked down and possible ETAs for fixes. That's not really ideal for a full price game, to be honest, and cause for a lot of frustration around various topics, imo.

Was thinking the same thing - this issue doesn't seem to have been properly registered nor documented in official channels. I was hoping that channels like DigitalFoundry would put the PC port in the limelight and force the team to acknowledge the issue, but that doesn't seem like it'll happen any time soon.

@holger1405Yeah, the performance is "fine" until you finish the mission there.

@PaxnosSeems like that area is possibly broken by a Denuvo related issue. I'm not well versed in these things, though. The area only goes to crap once you finished the mission there.

Has anyone else heard of/tried using this? https://github.com/Hin🤭okami/UE-Optimizer I'm using it (not sure the best settings regarding shaders), but it seems to reduce some of the noticeable hiccups in the game and makes the problematic spots observably smoother. To be clear, it doesn't eliminate or really lessen the major stuttering spots but they don't feel as erratic. To note, I'm also using SpecialK, including its framerate limiter.

Did some testing with and without using that optimizer at the two most troubling spots in the whole game.

With

Without

With

Without

Conclusion

My testing methods were not the best but going off the graphs it doesn't seem to make too much of a difference. Though at both locations, moving the camera feels smoother. As I said, it's perceptually better to me (I have a fixed refresh rate monitor) but it by no means fixes the problem.